Brain chirps: spectrographic signatures of epileptic seizures.

OBJECTIVE: A chirp is a brief signal within which the frequency content changes rapidly. Spectrographic chirps are found in signals produced from many biological and physical phenomena. In radar and sonar engineering, signals with chirps are used to localize direction and range to the signal source. Although characteristic frequency changes during epileptic seizures have long been observed, the correlation with chirps and chirp technology seems never to have been made. METHODS: We analyzed 19404 s (1870 s of which were from 43 seizures) of intracranially (subdural and depth electrode) recorded digital EEG from 6 patients for the presence of spectral chirps. Matched filters were constructed from methods in routine use in non-medical signal processing applications. RESULTS: We found that chirps are very sensitive detectors of seizures (83%), and highly specific as markers (no false positive detections). The feasibility of using spectral chirps as matched filters was demonstrated. CONCLUSIONS: Chirps are highly specific and sensitive spectrographic signatures of epileptic seizure activity. In addition, chirps may serve as templates for matched filter design to detect seizures, and as such, can demonstrate localization and propagation of seizures from an epileptic focus.

Dynamical behavior of autoassociative memory performing novelty filtering for signal enhancement.

This paper concerns the dynamical behavior, in probabilistic sense, of a simple perceptron network with sigmoidal output units performing autoassociation for novelty filtering. Networks of retinotopic topology having a one-to-one correspondence between input and output units can be readily trained using the delta learning rule, to perform autoassociative mappings. A novelty filter is obtained by subtracting the network output from the input vector. Then the presentation of a "familiar" pattern tends to evoke a null response; but any anomalous component is enhanced. Such a behavior exhibits a promising feature for enhancement of weak signals in additive noise. As an analysis of the novelty filtering, this paper shows that the probability density function of the weight converges to Gaussian when the input time series is statistically characterized by nonsymmetrical probability density functions. After output units are locally linearized, the recursive relation for updating the weight of the neural network is converted into a first-order random differential equation. Based on this equation it is shown that the probability density function of the weight satisfies the Fokker-Planck equation. By solving the Fokker-Planck equation, it is found that the weight is Gaussian distributed with time dependent mean and variance.